{
  "id": 11202584,
  "title": "The cost of AI performance is falling so fast it could reshape the entire software market",
  "url": "https://urgent.news/2026/10/01/the-cost-of-ai-performance-is-falling-so-fast-it-could-reshape-the",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-01T13:31:00.000Z",
  "source": {
    "name": "TechSpot",
    "slug": "techspot",
    "url": "https://www.techspot.com/news/114055-cost-ai-performance-falling-fast-could-reshape-entire.html"
  },
  "original_language": "en",
  "account": "AI models are becoming increasingly affordable, potentially reshaping the software market, according to a report from research firm Epoch AI. The study reveals that the cost of achieving a certain level of AI performance has decreased by nearly 50% every quarter, or about 13 times a year - a faster rate than the cost reductions seen in lithium-ion batteries, DNA sequencing, and computing hardware. This cost decline could render more capable models affordable for everyday use and make it difficult for AI developers to maintain pricing power. The report uses GPQA Diamond, a graduate-level multiple-choice test, as its main metric, measuring the cost to achieve a certain answer quality. A notable example is the price drop between OpenAI's GPT-3 and GPT-5.6 Luna, where the latter could reportedly achieve the same score for $0.0004 per question, down from $0.30 per question. While performance remains crucial for certain applications, the cost of delivering this quality is now equally important. When compared to other technological trends, AI's cost decline has been staggering - four times faster than DNA sequencing and 18 times faster than lithium batteries. However, the pace of price declines appears to be slowing, potentially due to the increasing expense of energy, data-center capacity, and AI hardware. Additionally, there's a risk of 'benchmaxxing,' where developers train models to perform well on benchmarks without translating to broader real-world improvements. Despite these challenges, the lower prices create more options for customers, potentially pressuring providers that invest heavily in building frontier systems. Nonetheless, factors like existing integrations, security policies, user habits, and trust also play a significant role in customers' decisions. The ultimate challenge for developers is creating products that remain useful and profitable even as the models behind them become increasingly affordable.",
  "summary": "Epoch's report looks at the cost of running models that meet a set performance level on difficult benchmarks. The analysis does not measure the average price of every AI product or task. Instead, it tracks what it costs to get a model to perform at a certain standard. Read Entire Article",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}